Using the ExEm unmixing method, it is possible to directly calculate the FRET efficiency of each of the cells in the microscope images to compare with reported values. In our case, however, an additional instrumental correction is required to allow pixels of all intensities to be included. To illustrate this, in Fig. 8 we show the FRET efficiency for each of the C5V cells, determined using the ExEm spectral unmixing method with no correction for the spectral sensitivity of the detectors. Clearly, at both the level of individual pixels [Fig. 8(a)] or within ROIs encompassing each cell [Fig. 8(b)], the calculated FRET efficiency changes dramatically as a function of the total brightness of the pixel or ROI. That is, fainter parts of the images have a lower FRET efficiency than brighter parts. A similar effect has previously been reported by Chen et al.27 and Levy et al.30 who noted that the PMT detectors on common microscopes do not respond linearly to different incident light intensities, and that this response may differ depending on the incident wavelength. In our case, the detectors show a much greater deviation from linearity in the red part of the spectrum (longer wavelengths) than in the blue (shorter wavelengths). As a consequence, the acceptor emission appears less intense relative to the donor emission in faint cells compared with bright cells as can be seen in the emission spectra plotted in Fig. 8(c). As can be seen in Fig. 8(a), the efficiency values approach a constant value at high intensities, so it is possible to gain accurate results by using only the brightest parts of the images. However, if reliable results are to be obtained independent of the pixel brightness, then a correction for the spectral response of the instrument is required. It is worth noting that while this nonlinear response is common in most PMT detectors, it may not be to the level reported here if the instrument has been well precalibrated and linearized. Furthermore, such nonlinearity is not generally present for CCD cameras which are more common on wide field microscopes. Thus the instrument spectral response correction may not be required in many cases.